diff --git a/README.md b/README.md
index bde83c5..8bbeac1 100644
--- a/README.md
+++ b/README.md
@@ -72,7 +72,7 @@ and engineering tips, tricks and best practices to build and train the neural ne
                      <a href="https://github.com/conversationai/conversationai-models/tree/master/attention-tutorial">code</a><br/><br/></div>
                      <p>Advanced RNN architectures for natural language processing. Word embeddings, text classification,
                      bidirectional models, sequence to sequence models for translation. Attention mechanisms. This session also explores
-                     Tensorflow's powerful seq2seq API. Applications: toxic comment detection and langauge translation.
+                     Tensorflow's powerful seq2seq API. Applications: toxic comment detection and language translation.
                      Co-author: Nithum Thain. Duration: 55 min</p></td>
     </tr>
     <tr>
diff --git a/tensorflow-mnist-tutorial/README_BATCHNORM.md b/tensorflow-mnist-tutorial/README_BATCHNORM.md
index f3b83cd..ba71b1a 100644
--- a/tensorflow-mnist-tutorial/README_BATCHNORM.md
+++ b/tensorflow-mnist-tutorial/README_BATCHNORM.md
@@ -13,7 +13,7 @@ Tensorflow has both a low-level and a high-level implementation for batch normal
 #### Low-level Tensorflow
 The low-level tf.nn.batch_normalization function takes your inputs, subtracts the average and divides by the variance
 that you pass in. It is up to you to compute both the batch statistics (average
-and variance of neuron outputs across a batch) and their moving averages across multiple batches and use them apropriately at trainig and
+and variance of neuron outputs across a batch) and their moving averages across multiple batches and use them appropriately at training and
 test time. It is also up to you to compute your batch statistics
 correctly depending on whether you are in a dense or a convolutional
 layer. Sample code is available in [mnist_4.2_batchnorm_convolutional.py](mnist_4.2_batchnorm_convolutional.py)
diff --git a/tensorflow-mnist-tutorial/mlengine/README.md b/tensorflow-mnist-tutorial/mlengine/README.md
index 4ea9ac6..73d88af 100644
--- a/tensorflow-mnist-tutorial/mlengine/README.md
+++ b/tensorflow-mnist-tutorial/mlengine/README.md
@@ -63,7 +63,7 @@ gcloud ml-engine local predict --model-dir checkpoints/export/Servo/XXXXX --json
 You can read more about [batch norm here](../README_BATCHNORM.md).
 
 If you want to experiment with TF Records, the standard Tensorflow
-data format, you can run this script ((availble in the tensorflow distribution)
+data format, you can run this script ((available in the tensorflow distribution)
 to reformat the MNIST dataset into TF Records. It is not necessary for this sample though.
 
 ```bash